Childhood bipolar disorder (BP-I) is a chronic illness that seriously impairs home, school and peer functioning and is associated with high rates of psychosis, mixed mania, ultradian rapid cycling and suicidality. Even bright children get failing grades at school and most require a myriad of medications. Given the newness of recognizing the existence of child BP, there is as yet scant knowledge of treatment, including whether medications that have shown effectiveness in adult BP will have similar effects in children. Regarding treatment, two of the more pressing issues have been (1) which class of drug to use first (lithium, valproate, atypical neuroleptics) and (2) which drug to add-on or switch to if the first drug fails. To investigate these research questions, the TEAM project was initiated two years ago as a complex, multi-site, multi-strata, equipoise stratification, adaptive strategy design to study 540 subjects with child BP-I (manic or mixed phase), aged 6-15. In this paradigm, subjects are randomized within three strata that include Stratum One for those on no study medications. Stratum Two is an add-on strategy for those who are partial responders to one study drug. Stratum Three is a cross-taper strategy for those who have failed one study medication. Also, subjects who are partial or poor responders in Stratum One are re-randomized to the other two strata. Due to the novelty and cost of this design, NIMH requested that TEAM begin as a two year pilot, with the provison that if the two year pilot demonstrated feasibility, the investigators would be encouraged to apply for an additional four years of funding to complete the study. At the time of this writing (5/27/04 for a 6/7/04 submission date), there are 107 randomized subjects, which exceeds the planned recruitment rate. Furthermore, all three strata are being filled. Therefore, feasibility has been established. As a large cooperative agreement, this study is monitored on a quarterly basis by the NIMH Data and Safety Monitoring Board (DSMB). At the five DSMB audits conducted at the time of this writing, TEAM has received outstanding praise for data management, quality control, and safety. The structure of this cooperative agreement is one coordinating site (Washington U in St. Louis, PI: B. Geller) and five data collection sites (Children's National Medical Center, PI: P. Joshi; Johns Hopkins Medical Institutions, PI: J. Walkup; U Pittsburgh, PI: D. Axelson; UT Medical Branch, Galveston, PI: K. Dineen Wagner; Washington U in St. Louis, PI: J. Luby). Dr. Ben Vitiello is the NIMH scientific collaborator. The chief study statistician, Dr. Satish lyengar is subcontracted to the coordinating site at Washington U in St. Louis and Dr. Philip Lavori is a major consultant for statistical analyses.